Like every other person in a creative field, I’ve been forced to consider what I’m going to do with my craft in a world with AI.
A couple of weeks ago I had a bit of a dark night of the soul about it. My annual time lord, Venus, conjoined Pluto in my tenth house of career and public life.
I started writing this piece in response to the thoughts and emotions that bubbled up during that transit. I continued to ruminate on it over the proceeding days as a number of other planets – the Moon, Sun, Mercury and Mars – all marched out of Capricorn’s cold stone hallways and strode into the neon-flooded Aquarian streets, passing Pluto along the way.
Before I continue this essay, allow me to wax poetic for a moment:
A shadowy figure lurks in a recessed doorway. Neon lights reflect in the figure’s sunglasses, obscuring the eyes and casting back a lurid impression of the viewer’s own gaze.
A surge of fear.
Adrenaline.
Crackling nerves.
Each member of the quintet passes that dark silhouette in turn.
Each locks eyes and feels their self temporarily lost in neon haze.
Each is captured, broken.
At the pulsing core of that singularity, certainty rises: an affirmation of who they are,
who they were,
who they will be.
Unmade and made again, over and over.
Blindly they stumble past.
With distance the terror subsides.
With distance the knowing fades too.
Only in the half light of pre-dawn can they remember that bolt of truth,
a dark pearl shining amidst clutching shadows.
When AI first became a thing a couple years ago, my first reaction was a knee-jerk Luddite response.
I didn’t want anything to do with AI, thank you very much. I had no interest in messing around with LLMs like ChatGPT. They repulsed me on a fundamental level, probably because I’m a writer. A writing machine felt like a direct threat to who I am. It gave me all sorts of icky feelings, so I opted to mostly tune it all out.
Note: I use “AI” and “LLMs” somewhat interchangeably in this post. I mainly refer to “LLMs” (Large Language Models) as I’m discussing writing-focused AI software, though I also use “AI” when talking about Artificial Intelligence broadly.
Then it came up in my corporate day job. Our managers encouraged us to use an AI tool (Microsoft Copilot) to be more “efficient.” I grudgingly tried it out a couple times and found the process frustrating and useless.
Copilot was either unable to do what I needed, or I still had to check its work for accuracy. In all cases, using Copilot took me longer than if I had just done the original task myself.
We were also urged to have Copilot write/edit our emails to make them more friendly! Or professional. Or a mixture of both friendly! and professional.
Yeah, no thanks. I have a fucking Honours English degree. I can write my own damn emails.
More recently, I started considering AI in relation to my creative work, writing about astrology. This is where the existential dread and despair started to sink in.
I played around with ChatGPT to generate astrology horoscopes and laughed at the garbage it generated. It got the dates of important astrological transits wrong, and the interpretations were nothing more than a pile of Twelve-Letter-Alphabet slop.
Ha, I thought. Hahaha. Vindication! An LLM can’t do astrology better than me.
Then I started reading some other astrologers’ work on Substack and realized a lot of them were using LLMs to write some or all of their articles. It was pretty easy to figure out if the writer was using an LLM, because it felt a lot like eating a bag of potato chips: it was easy to consume and I thought liked it well enough, but shortly afterwards I felt kind of gross.
I decided to keep reading these posts, however. I had avoided the AI issue until now, but I felt like I should experience more of it to see if my concerns were truly warranted.
In a very short time (no more than a couple of days), I couldn’t take it anymore. The posts written by an LLM were so obvious and so lacking. While they might seem passable with a quick skim, they really weren’t providing any deep or particularly meaningful insights into astrology. They rang with the bland hollowness and glossy mediocrity that’s the hallmark of AI output.
But what really got to me was the sheer quantity of content. Some accounts were pumping out articles every day, often several. My inbox was quickly flooded.
Instead of just shrugging and unsubscribing, however, I spiralled into existential despair.
What’s the point of continuing to write and share my articles, I mused, when all these other people are using LLMs to post dozens of articles in half the time that it takes me to write one?
What’s the point of spending hours researching and writing something, only to have ChatGPT and other LLM bots scrape my work in an instant and regurgitate it for someone else to post as their own?
Will readers know – or care – if a human wrote these words, versus an LLM?
For a couple of days, I was more discouraged with my astrology work than I’ve been in a long time. Maybe ever. It just felt pointless.
What got me out of it was doing some astrology. Go figure.
You’d think that us star gazers would be in tune with the current transits all the time. It seems like we should all be ready and waiting for any difficulties to pop up, because we have the wonderful gift of foresight that comes from deep knowledge of these cycles and energies.
Yeah, no. We get blindsided all the time.
Once I stopped feeling sorry for myself long enough to check the transits, I was able to gain some perspective.
But I still had to make a decision about how I was going to move forward in my astrology work. I had three choices:
- Double down on ignoring AI/LLMs and just keep chugging away on my own thing.
- Give in and start using LLMs.
- Give up and stop writing completely.
Even at my darkest moment, Option #3 was never really an option.
I’ve always been a writer. Astrology is what I write about now, but before that it was wine and food and theatre. I need to write. It’s like a reflex. Or a compulsion.
Writing is also how I process information and make sense of the world. My blog has always been the site (heh) of this, so I’m essentially learning in public every time I post an article. This is something I can’t stop doing. My curiosity is too strong.
I could stop making these essays public, but I think it’s fun to share everything I’ve learned. I hope it helps others who join me on this journey, whether they just dip in for a moment after finding my article through a random Google search, or whether they have been following me for years. (Hello to all of you, and thanks for reading!)
As for astrology, I truly feel called to it. That might sound cheesy or weird, but a few years ago the planets really did tell me this was my path.
For those curious: it was the 2020 Saturn–Pluto conjunction in Capricorn, within a degree of my Midheaven. Believe me when I say I had no choice in the matter.
Furthermore, my motivation was never to pump out as much content as possible to game an algorithm for a quick buck. I have the luxury of not relying on astrology for my main income, so the financial pressure doesn’t hit me the way it does many other writers and creatives.
I understand the impulse, and the current panic that this technology has caused. But I truly believe no one is going to get rich using AI tools to pump out as much slop as possible in a desperate attempt to stay relevant. You’re just spamming our inboxes and social media feeds, and eventually we are all going to unsubscribe and mute you.
Once I established that I wasn’t going to stop writing, I thought I’d go for Option #1. Fuck AI, I thought. I’m human, I’m a writer, I’ll just keep doing my thing.
But…I kept hearing about how great a tool these LLMs could be, and wondering if I was missing out. And then, an article by Sadalsvvd landed in my inbox.
Sadalsvvd has posted some excellent essays on astrology and other esoterica. A while back, I went on my friend Mat Dragonstone’s podcast, along with fellow astrologer Zamboni Funk, to talk about one of these pieces, “Astrology Has a Scarcity Mindset Problem.” That was a great discussion – here’s the link if you missed it.
Sadalsvvd’s latest essay isn’t about astrology directly, but rather is an excellent and extremely comprehensive resource that intelligently and fairly addresses all the major issues and complaints about AI/LLMs – environmental, social, philosophical and more. I’ll be referencing it repeatedly.
It also talks about how LLMs can be a liberating technology. OK, you have my attention.
Now, Sadalsvvd works in software engineering and therefore has a front-row seat to the best and worst that LLMs can do. They’ve personally tested all the major LLMs out there, and seem generally very optimistic about this technology.
What really stood out for me was that while people can and will use LLMs to do the writing for them, Sadalsvvd’s opinion is that they are better used to handle all the crappy tasks that suck to do. Here’s a direct quote from the article:
“My rule is: if there’s a task that requires you to use language to understand and complete it, but is not actually interesting or rewarding to do, an LLM can and probably should handle it for you. You have more important things to do with your time!”
The examples given in that article are things that will benefit me most in my corporate day job. But it got me wondering about possible applications in my creative astrology work.
Now, I’m resolutely against using an LLM to write for me. This is never going to change, no matter how good the technology gets. I write because I like writing. Why would I outsource that?
However, there are some tedious tasks involved in astrology work. I am reminded of the fact that it wasn’t very long ago when astrologers had to calculate charts by hand. While I think this is a great skill to have (and one I plan to learn, though I haven’t just yet), I don’t have a problem using software to do this work for me. So what other tasks could be outsourced without losing any integrity in my work?
Mundane astrology requires a lot of historical research, in order to spot astrological trends and thereby gain insight into what might come in the future.
I like doing this research but my time is limited. I can’t spend all day studying history (alas). I’m curious about whether I could successfully use an LLM to compile lists of historical events alongside their dates and key astrological transits. This will require diligent fact-checking, but that’s something I have to do for myself anyway. LLMs hallucinate; humans make typos. Both need a good editor.
So, while I’m extremely skeptical about the quality of work that an LLM could do with regards to astrological research, I’m not completely opposed to trying it out. It’s worth a shot just to see what it can and can’t do.
My initial experiments using ChatGPT for historical research haven’t blown me away, but they weren’t so bad as to be useless. I will keep trying it out to see if it is actually worth relying on an LLM to do some of this legwork for me.
Other than that, I’m intrigued by the idea of using an LLM to help me redesign my blog. I loathe blog design with every fibre of my being. I would be thrilled to outsource it. That said, this is nowhere near the top of my priority list, so I’m not sure when I’ll test that one out.
So, it turns out I’m not a complete Luddite. I’m still extremely cautious about how, where, and why I might use an LLM.
But I can’t escape the fact that this technology is clearly here to stay. The Uranus-Pluto trine is upon us and each of us is going to have to come to terms with how we will use it.
One thing I want to note before I end this meandering post: I haven’t addressed the issue of AI-generated images, music or other forms of non-written creative work. My stance on those is a lot faster to articulate: it’s all garbage.
I have yet to find an AI-generated image that didn’t look weird and uncanny at best, or horrifying slop at worst. I’m sure there are some exceptions, but those are vanishingly rare.
Copyright infringement is also a much bigger issue with AI generated imagery as opposed to LLM generated writing (which still has issues, but not as much). Sadalsvvd sums it up eloquently in the first footnote of the LLM essay, noting that generative image models have training sets that “almost universally violate copyright rights and creative ownership laws and are used extensively for disinformation, revenge pornography, and CSAM material. I am hard-pressed to find actual ethically justifiable deployments of image models, especially in business.”
I’d like to point you to my friend Vanessa’s recent piece about the use of AI images in contemporary occulture for a great discussion on the importance of aesthetics in magic work. Astrology writing isn’t magic, but it’s certainly not enhanced by slapping a crappy AI-generated image on it. And from what I see online, astrologers are some of the worst offenders on the issue of bad aesthetic taste with regards to AI imagery.
I try my best not to use any AI-generated imagery on my blog. Weird, cool or silly photoshopped images, sure. But not the AI stuff.
Before I end this post, I want to make one final thing very clear: I will never use an LLM to write my blog posts. Not now; not ever.
And just because you see em dashes in my writing doesn’t mean I’m using an LLM. I was using those for years before ChatGPT came along. They’re mine and I’m not giving them up.
Thanks for reading this meandering, navel-gazing post. I hope you got something out of it and I know I’m not alone in thinking through these issues. AI literacy seems like it’s going to become a core competency in very short order, so we’re all going to have to find ways to engage – or refuse to engage – with it, intelligently and meaningfully.
If nothing else, I want to be able to look back at this post in one year, three years, five years, twenty years, and laugh at how naïve I was.